Startup Luxonis has officially launched its crowdfunding campaign for DepthAI, a depth-sensing computer vision development board available in Raspberry Pi HAT, Compute Module, and universal USB 3.0 flavors — as well as a system-on-module (SOM) for integration into finished designs.
First unveiled two months ago, and originally designed to use a neural network inference engine combined with depths sensing and object tracking to improve cycle safety, Luxonis' DepthAI is based around the Intel Movidius Myriad X vision processing unit (VPU). A successor to the company's earlier Movidius 2 VPU, the Myriad X includes a Neural Compute Engine (NCE) hardware accelerator, 16 programmable Streaming Hybrid Architecture Vector Engine (SHAVE) cores, and — key to the DepthAI design — vision processing hardware including a stereo depth-sensing block which can process dual 720p video feeds at refresh rates of up to 180Hz
"DepthAI is a platform built around the Myriad X to combine depth perception, object detection (neural inference), and object tracking that gives you this power in a simple, easy-to-use Python API," Luxonis explains of its design. "It's a one-stop shop for folks who want to combine and harness the power of AI, depth, and tracking. It does this by using the Myriad X in the way it was intended to be used — directly attached to cameras over MIPI — thereby unlocking power and capabilities that are otherwise inaccessible.
"[DepthAI] is engineered from the ground up, with custom hardware, firmware, and an elegant, easy-to-use Python interface that allows you to easily leverage this insane power. We've designed our DepthAI product family to be easy to use."
Each DepthAI board includes real-time object identification, location, and 3D tracking at 25 frames per second. "The easiest-to-use setup is our DepthAI Raspberry Pi Compute Module version, which boots up running Object Localisation on 20 classes (PASCAL VOC 2012)," Luxonis notes. "Need to know where people are and what they're doing? Download and run the person detection model. Need to know the location of all the honey bees? Download and run a bee-count model. Or, if you need something custom, you can train your own models based on available/public datasets and then use OpenVINO to deploy them to DepthAI."
The company is launching three key versions of the DepthAI platform, all based on a $99 system-on-module (SOM): The Raspberry Pi HAT Edition adds a DepthAI module to a Raspberry Pi 3 Model B and Model B+ or Raspberry Pi 4 using the Hardware Attached on Top (HAT) standard; the Raspberry Pi Compute Module version is a standalone board using the SOM version of the Raspberry Pi; and the USB 3.0 variant offers OpenVINO compatible acceleration over a USB connection.
Luxonis has set pricing for its crowdfunding campaign at $99 for the SOM, $149 for the USB 3.0 or Raspberry Pi HAT variants, and $299 for the Raspberry Pi Compute Module version — the latter including the Compute Module required to get up and running. The company is also offering a 4K-resolution 60Hz camera module or pair of 720p 120Hz camera modules at an additional $74. All items are expected to ship in mid-February 2019.
More information is available on the DepthAI Crowd Supply campaign page.